Cyber threats are growing increasingly adaptive, demanding reasoning-driven approaches to both defense and offense. This work presents a suite of agentic AI systems that leverage Large Language Models (LLMs) to automate critical cybersecurity workflows across defensive and offensive domains. On the defensive side, MalAware automates comprehension of malicious software behaviors by generating human-readable explanations from ground-truth malware analysis data. PARAG (Proactive Answering framework integrating LLMs with Retrieval-Augmented Generation) automates security-policy question answering and compliance tasks. AURA (Attribution Using Retrieval-Augmented Agents) applies an agentic RAG framework to threat-intelligence attribution, reaching up to 100% top-2 nation-wise attribution accuracy with Claude 3.5 Sonnet on a benchmark of over 2,200 threat reports. On the offensive side, MalGEN employs a multi-agent generative pipeline to synthesize behaviorally diverse, ATT&CK-aligned malware artifacts across eight LLMs and two platforms, producing a 1,920-artifact benchmark and exposing detection blind spots in conventional antivirus pipelines. For Capture-The-Flag (CTF) style offensive tasks, D-CIPHER introduces a dynamic multi-agent system pairing a planner with heterogeneous executors, while CRAKEN augments LLM agents with structured knowledge retrieval for exploit execution. Finally, a systematic study of hyperparameter tuning (temperature, top-p, max tokens) for offensive-security agents introduces CTFTiny, a lightweight 50-task CTF benchmark, and CTFJudge, an LLM-as-a-judge evaluator that scores agent trajectories against expert solutions for fine-grained, beyond pass/fail assessment. Together, these systems show how agentic LLM reasoning can automate and scale both attack and defense workflows in modern cybersecurity.

Prof. Sandeep Kumar Shukla is the Director of the International Institute of Information Technology, Hyderabad (IIIT-H). He earned his Ph.D. in Computer Science from the University at Albany in 1997, following an M.S. from the State University of New York (1995) and a B.E. in Computer Science and Engineering from Jadavpur University, Kolkata (1991).Before joining IIIT-H, he was Rajiv & Ritu Batra Chair Professor in Cyber Security and Professor (HAG) of Computer Science and Engineering at IIT Kanpur, where he headed the CSE department from 2017 to 2020 and held the Poonam & Prabhu Goel Chair Professorship from 2016 to 2019. He was also founder and joint coordinator of the National Interdisciplinary Centre for Cyber Security & Cyber Defence of Critical Infrastructures (C3i Centre).

His earlier academic affiliations include MIT, Virginia Tech, Inria (France), and TU Kaiserslautern.He has authored over 300 peer-reviewed papers and held editorial roles with leading IEEE and ACM journals. His honours include IEEE Fellow (2014), ACM Distinguished Scientist (2013), the Humboldt Bessel Award (2008), and the US Presidential Early Career Award for Scientists and Engineers (2004).
He advises RBI's ReBIT, SEBI, NPCI, IRDAI, and the Ministry of Power.